On 28 April 2011 22:56, Luke <hazelnusse@gmail.com> wrote:
I am thinking that perhaps the approach I should be taking should
involve contouring the real part of the eigenvalues which determine
the stability, and then plot the zero-level curve.  I'll have to think
about that some more.

This sounds like a very sensible approach and is quick and easy to try out using tricontour/tricontourf.  You may have to use a very small positive value for the contour level rather then zero to get what you want.
Is it clear what I am trying to do?  If so, do you think the Delaunay
triangulation is the right way to go?

Yes, it is clear what you are trying to do.  I think that you shouldn't be concerned with the triangulation, Delaunay or not, as this is at too low a level for what you are attempting.  Stick to the high-level data analysis and presentation functions like tricontour and ignore details of the underlying triangulation.  If you are having to manipulate a triangulation then you are becoming a computational geometrist - it is a completely valid and interesting thing to do but is probably taking your attention away from your real work.